Energy-Efficient Resource Allocation with Dynamic Cache Using Reinforcement Learning

With the increasing amount of data in wireless network, the problem of energy consumption becomes more serious due to energy requirement for massive data transmission. This paper proposes an energy-efficient resource allocation algorithm with dynamic cache, which can adjust the caching strategy dynamically according to the channel state to reduce energy consumption under the constraint of smooth video streaming. The mathematical models of energy consumption for video transmission and decision selection are established, respectively. Given the dynamic channel environment, an on-line algorithm using reinforcement learning is proposed. In order to reduce the overall energy consumption of the system, and maintain the balance of energy consumption between transmission and calculation, the model of the off-line part is trained by using the neural network, and the calculation accuracy is adjusted adaptively. The simulation results show that the proposed algorithm can improve the total energy efficiency of the system effectively.